JP7394713B2 - Melting furnace monitoring system, melting furnace monitoring method, and program - Google Patents

Description

The present invention relates to a melting furnace monitoring system, a melting furnace monitoring method, and a program.

Patent Document 1 discloses that a slag region is extracted from each of a plurality of consecutive images of a slag outlet of a melting furnace by determination based on a brightness threshold, and an area where a slag region is commonly extracted in the plurality of images is extracted. A method is disclosed for specifying a fixed slag region.

Japanese Patent Application Publication No. 2019-152652

By the way, since the amount and properties of slag flowing through the slag outlet vary depending on the equipment, a situation may occur in which slag continues to flow in the same place depending on the equipment. In the above method, there is a possibility that a portion where slag continues to flow in the same place may be determined to be a fixed slag region.

The present invention has been made in view of the above problems, and its main purpose is to provide a melting furnace monitoring system, a melting furnace monitoring method, and a program that can improve the accuracy of determining fixed slag at the slag outlet. It is about providing.

In order to solve the above problems, a melting furnace monitoring system according to one aspect of the present invention includes a camera that continuously photographs the slag outlet of a melting furnace to generate a plurality of time-series images, and a determination based on a brightness threshold. a slug candidate area extraction unit that extracts slug candidate areas in the plurality of images; a motion detection unit that detects an area in which movement has been detected in the plurality of images as a motion detection area; A fixed slug area specifying unit that specifies an area other than the motion detection area as a fixed slug area.

Further, in a melting furnace monitoring method according to another aspect of the present invention, a camera continuously photographs the slag outlet of the melting furnace to generate a plurality of time-series images, and the plurality of images are determined by a brightness threshold. extracting a slug candidate region in the plurality of images, detecting a region in which movement is detected as a motion detection region, and identifying a region of the slag candidate region excluding the motion detection region as a fixed slag region. do.

Further, a program according to another aspect of the present invention includes an image acquisition unit that acquires a plurality of time-series images generated by a camera that continuously photographs the slag outlet of a melting furnace; a slug candidate area extraction unit that extracts a slug candidate area in an image; a motion detection unit that detects an area in which movement is detected in the plurality of images as a motion detection area; and a motion detection unit of the slug candidate area. The computer is made to function as a fixed slag area specifying section that specifies the area excluding the area as a fixed slag area.

According to the present invention, it is possible to improve the accuracy of determining fixed slag at the slag outlet.

1 is a diagram illustrating a configuration example of a melting furnace monitoring system according to an embodiment. FIG. 2 is a diagram illustrating a configuration example of a processing device. FIG. 3 is a diagram showing an example of an image taken by a camera. 1 is a diagram illustrating an example of a procedure of a melting furnace monitoring method according to an embodiment. It is a figure which shows the specific example of a procedure of fixed slag area|region acquisition process S4. It is a figure following FIG. 5. It is an explanatory diagram of S3. It is an explanatory diagram of S4-5a. It is an explanatory diagram of S4-5a. It is an explanatory diagram of S4-5b2 and 5b3. It is an explanatory diagram of S4-5b5 and 5b6. It is an explanatory diagram of S4-5b7. It is an explanatory diagram of S4-6.

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a diagram showing a configuration example of a melting furnace monitoring system 100 according to an embodiment. The melting furnace monitoring system 100 is a system for monitoring the gasification melting furnace 200. The melting furnace monitoring system 100 includes a processing device 1 and a camera 3.

The gasification and melting furnace 200 is a gasification and melting furnace used, for example, in a garbage treatment facility. The gasification melting furnace 200 includes a melting furnace 21 . Combustible gas and unburned components thermally decomposed in a gasification furnace (not shown) flow into the melting furnace 21 through the inlet passage 23 . The melting furnace 21 generates slag by burning unburned matter at high temperature and melting the ash. The molten slag is discharged from a slag outlet 25 formed at the lower part of the melting furnace 21 and taken out of the melting furnace 21 through a discharge passage 27 .

A camera 3 for photographing the slag outlet 25 is provided below the melting furnace 21 . The camera 3 is arranged diagonally below the slag outlet 25 and photographs the slag outlet 25 through a monitoring hole 29 provided in the discharge path 27. The camera 3 continuously photographs the slag outlet 25 to generate a plurality of time-series camera images, and outputs them to the processing device 1. The plurality of time-series camera images may be, for example, a plurality of still images (frames) included in video data, or a plurality of still images individually generated by shooting at predetermined time intervals. good.

The processing device 1 identifies the fixed slag stuck to the slag outlet 25 based on the camera image supplied from the camera 3, and calculates the aperture ratio of the slag outlet 25. The calculated aperture ratio data is presented to the operator P or transmitted to the control device of the gasification and melting furnace 200. The operator P who has referred to the aperture ratio data or the control device that has received the data executes various operations to improve the aperture ratio of the slag outlet 25.

FIG. 2 is a block diagram showing an example of the configuration of the processing device 1. As shown in FIG. The processing device 1 is a computer including a CPU, RAM, ROM, nonvolatile memory, input/output interface, and the like. The CPU of the processing device 1 executes information processing according to a program loaded into the RAM from the ROM or nonvolatile memory.

The program may be supplied, for example, via an information storage medium such as an optical disk or a memory card, or may be supplied, for example, via a communication network such as the Internet or a LAN.

The processing device 1 includes an image acquisition section 11, a slag candidate region extraction section 12, a positional deviation correction section 13, a motion detection section 14, a motion detection region correction section 15, a fixed slug region identification section 16, and an aperture ratio calculation section 17. ing. These functional units are realized by the processing device 1 executing information processing according to a program.

FIG. 3 is a diagram showing an example of a camera image taken by the camera 3. As shown in FIG. Within the range of the slag outlet shown in the camera image, an opening through which the inside of the furnace can be seen, molten slag, and fixed slag appear. Among these, molten slag and fixed slag have close brightness and are difficult to distinguish.

In the method of Patent Document 1, areas where slag candidates are commonly extracted in consecutive images are determined as fixed slag, but depending on the equipment, a situation may occur where slag continues to flow in the same location. Therefore, in this method, such a portion may be determined to be a fixed slag, and the accuracy of calculation of the aperture ratio may decrease.

Therefore, in this embodiment, as will be explained below, the accuracy of determining fixed slag is improved by focusing on the movement of the slag surface.

FIG. 4 is a diagram illustrating a procedure example of the melting furnace monitoring method according to the embodiment, which is implemented in the melting furnace monitoring system 100. FIGS. 5 and 6 are diagrams illustrating a specific procedure example of the fixed slag area acquisition process S4. The processing device 1 executes the information processing shown in FIGS. 4 to 6 according to the program.

First, the processing device 1 acquires the camera image Xcam(t) at time t, performs preprocessing, and obtains the image Xrn(t) (S1: processing as the image acquisition unit 11). Since the subsequent processing assumes the use of grayscale images, the camera image is converted to grayscale in the preprocessing.

As the preprocessing, noise removal processing or the like may be added. Examples of noise removal processing include a median filter, average filter, gaussian filter, biliteral filter, open filter, and close filter. Additionally, if the camera image is an interlaced image, deinterlacing processing (obtaining only data in even columns and rows or odd columns and odd rows) may be performed, and if the camera image is an analog image, electrical noise The striped pattern generated due to this may be detected from the projected brightness distribution for each row or column, and then corrected.

Next, the processing device 1 identifies the position of the slag outlet using the image Xrn(t) obtained in S1 above, and obtains an image Xana(t) corrected so that the center of the slag outlet area is located at the center of the image. (S2).

Specifically, <Processing 1> A template matching process is performed on Xrn(t) using the pre-registered slag opening shape mask image M as a template, or <Processing 2> A position correction process of the slag opening is performed. The slag outlet region is detected from the image Xrn(t) by performing template matching processing on Xrn(t) using the image Xana(t-1) at the previous time as a template. In this embodiment, Process 1 is performed every time a certain period of time Tspan has elapsed, and Process 2 is performed at a timing when Process 1 is not performed. Thereafter, the image is translated in parallel so that the center of the circumscribed rectangle of the detected slag outlet (the diagonal intersection of the circumscribed rectangle) coincides with the diagonal intersection of the image, and a corrected image Xana(t) is obtained.

Next, the processing device 1 obtains the fixed slag candidate region S(t) at time t based on the image Xana(t) obtained in S2 above (S3: Processing as the slag candidate region extraction unit 12) . FIG. 7 is a diagram showing an example of (a) the analysis source image Xana(t) and (b) the fixed slag candidate region S(t). In the figure (b), the white portion represents the fixed slag candidate region S(t).

In this process, as in Patent Document 1, a fixed slag candidate region S(t) is extracted by determination based on a brightness threshold. Specifically, the brightness unevenness of the camera image is suppressed, the brightness abruptly changing part of the camera image is extracted, the brightness histogram is calculated by excluding the brightness suddenly changing part from the camera image, the threshold value is automatically determined based on the brightness histogram, and the second brightness change part is extracted. A value conversion process is executed to extract a region whose brightness is less than or equal to a threshold value as a fixed slag candidate region S(t).

Next, the processing device 1 executes fixed slag area Y(t) acquisition processing (S4). In the fixed slag region Y(t) acquisition process, fixed slag and molten slag are distinguished from the fixed slag candidate region S(t) using motion detection of the molten slag, and the fixed slag region Y(t) is obtained. Here, a portion of the stuck slag candidate region S(t) that continues to be detected without movement for a certain period of time is determined to be a stuck slag region Y(t).

The fixed slag area Y(t) acquisition process S4 will be explained in detail using the flowcharts of FIGS. 5 and 6. In this embodiment, three ring memories (RM1, RM2, RM3) of length N are prepared for performing the processing.

As shown in FIG. 5, the processing device 1 first sets the evaluation time length Tt1 according to the elapsed time from the start of the process (S4-1). Specifically, if the elapsed time Tp from the start of the process is less than the set value, the evaluation time length Tt1 is set to the evaluation time length Tt1, and in other cases, the evaluation time length Tt1 is set to the set value. Let it be N. This is a process for determining the number of data to be processed when sufficient data is not stored in the ring memory.

Next, the processing device 1 obtains the analysis source images Xana(t) and Xana(t-1) at time t and time t-1, and the stuck slag candidate region S(t) at time t, and The candidate area S(t) is recorded in the ring memory RM1 (S4-2, 3).

Next, the processing device 1 determines whether the elapsed time Tp from the start of the process exceeds the set value Th1 (S4-4). This is a process for determining whether a sufficient number of images (T1 or more images) are stored in the ring memory RM1.

If the elapsed time Tp from the start of the process exceeds the set value Th1 (S4-4: YES), the processing device 1 performs the following two processes using the images from time t-Tt1 to time t. That is, processes S4-5a1 to S4-5a1 to 3 for finding a common area B(t) that has continued to be a fixed slag candidate area S(t), and processes S4-5b1 to S4-5b1 for calculating a region E(t) where movement of molten slag has been detected. 7.

On the other hand, if the elapsed time Tp from the start of the process does not exceed the set value Th1 (S4-4: NO), the processing device 1 processes the stuck slag candidate area recorded in the ring memory RM1 from time t-Tt1 to time t. The area common to all images of S(t-Tt1) to S(t) is extracted using a logical product operation (AND operation) and set as a common area B(t) (S4-5c1), and the common area B(t ) is output as the fixed slag area Y(t) (S4-9b).

Processing S4-5a1 to S4-5a3 will be explained. In this process, the processing device 1 finds a common area B(t) that continues to be the fixed slag candidate area S(t) (processing as the slag candidate area extraction unit 12). The same process as in Patent Document 1 may be performed, but here, in order to further improve the accuracy of the aperture ratio calculation, detection omissions in the stuck slag candidate region S(t) at each time are corrected (S4-5a1).

Specifically, first, the processing device 1 derives the missed detection area A(t-1) of the stuck slag candidate area S(t-1) at time t-1 recorded in the ring memory RM1 from the following equation. (S4-5a1).

A(t-1)=~S(t-1) and (S(t) and S(t-2))

Here, and is a logical product, and ~ is an operator that inverts a black and white image.

Next, the processing device 1 corrects the stuck slag candidate region S(t-1) using the following equation to remove the missed detection of the stuck slag candidate region S(t-1) (S4-5a2).

S(t-1)=S(t-1) or A(t-1)

Here, or is a logical sum.

FIG. 8 is an explanatory diagram of steps S4-5a1 and S4-5a2. In the figures (a) to (c) and (e), the white portion represents the stuck slag candidate region S(t). In the figure (d), the white part represents the detection failure area A(t-1).

Finally, the processing device 1 performs a logical product of all the images of the fixed slag candidate areas S(t-Tt1) to S(t) from time t-tT1 to time t, reflecting the above correction, and (t) is obtained (S4-5a3). FIG. 9 is an explanatory diagram of step S4-5a3. Through the above processing, the common area B(t) of the fixed slag candidate area S(t) can be determined with high precision.

Processes S4-5b1 to S4-5b1 to S4-5 will be explained. In this process, the processing device 1 determines the area E(t) where the movement of the molten slag is detected. That is, this process calculates the logical sum of the regions in which the pattern on the slag surface observed from time t-tT1 to time t has moved, and considers that region as molten slag. Specifically, the following steps are performed.

First, the processing device 1 obtains an image Xana'(t) whose positional deviation with respect to the immediately previous frame has been corrected (S4-5b1: processing as the positional deviation correction unit 13). Specifically, using the image Xana(t-1) at time t-1 as a template, template matching is performed on the image Xana(t) at time t. As a result, a shift between images is detected, and image Xana(t) is corrected to obtain image Xana'(t). This processing is performed to eliminate the influence of camera vibration when it occurs.

Next, the processing device 1 detects the displacement amount of the movement of the molten slag by optical flow (S4-5b2: processing as the movement detection section 14). Specifically, optical flow is executed using the image Xana(t-1) at time t-1 and the image Xana'(t) at time t as input, and the displacement of the movement of the molten slag from time t-1 to time t is calculated. Find the quantity. Optical flows include sparse optical flows and dense optical flows, and in this embodiment, it is preferable to use dense optical flows.

Next, the processing device 1 extracts a region where the displacement amount is a certain value or more by applying threshold processing to the determined displacement amount, and obtains a motion detection image C(t-1) (S4-5b3 : Processing as the motion detection unit 14). FIG. 10 is an explanatory diagram of steps S4-5b2 and 5b3. In the diagram (c), the white portion represents an area where the amount of displacement is equal to or greater than a certain value. Note that as a motion detection method, not only optical flow but also a background subtraction method that performs motion detection by statistically estimating a background image from a plurality of frames may be used.

Next, the processing device 1 removes noise from the motion detection image C(t-1), obtains a motion detection image C'(t-1), and records it in the memory (S4-5b4). Noise can be removed using, for example, fine motion detection leakage removal (Close filter), motion detection area blur removal (Erode filter), fine motion detection area removal (Open filter), or area expansion of A(t-1). removal, etc. This is done because the results of motion detection using optical flow include over-detection or omission of detection, and the shape detection accuracy of moving objects is low.

Next, the processing device 1 expands the motion detection area in the image C'(t-1) within the fixed slug candidate area S(t-1) to obtain an image D(t-1), and It is recorded in RM2 (S4-5b5, 6: processing as the motion detection area correction unit 15). FIG. 11 is an explanatory diagram of steps S4-5b5 and 5b6.

Specifically, (1) the motion detection area C(t-1) is expanded by a radius r' (a value smaller than the minimum area width of the slug), and (2) the motion detection area C(t-1) after expansion is From this, the overlapping region with the fixed slag candidate region S(t-1) is deleted (to prevent expansion beyond the fixed slag candidate region). (3) The processes of (1) and (2) above are repeated until the maximum n that satisfies r'×n≦radius r. (4) Finally, after expanding by the radius (r-(r'×n)), perform the process in (2) above.

Since the noise removal process in S4-5b4 is highly effective, the motion detection area C'(t-1) is smaller than the area where the molten slag is actually moving. Therefore, correction is performed by applying expansion processing. At this time, by expanding according to the steps (1) to (4) above, the motion detection area is prevented from expanding beyond the fixed slug candidate area. Thereby, the motion detection area can be corrected based on the shape of the slug.

Next, the processing device 1 calculates the logical sum of the motion detection images D(t-Tt1) to D(t-1) at times t-tT1 to t-1 recorded in the ring memory RM2, and performs motion detection. An integrated image E(t) is acquired (S4-5b7). FIG. 12 is an explanatory diagram of step S4-5b7. In this way, by extracting a region in which movement is detected in any of the images at a plurality of times as a motion detection integration region, a motion region of the molten slag can be detected with high precision.

Next, the processing device 1 calculates the difference F(t) between the common area image B(t) of the stuck slug candidate and the motion detection integrated image E(t), and records it in the ring memory RM3 (S4-6: processing as the slag area specifying unit 16). Specifically, the difference F(t) is calculated using the following equation.

F(t)=B(t) and ~E(t)

Here, and is a logical product, and ~ is an operator that inverts a black and white image.

FIG. 13 is an explanatory diagram of step S4-5b7. In the figure (a), the white part represents the common area of fixed slag candidates. In the figure (b), the white part represents the motion detection integration area. In the diagram (c), the white portion represents the difference area. This process is a process of excluding the motion detection integration area from the common area of stuck slug candidates. That is, this is a process of extracting a region where a fixed slug candidate region in which no movement is detected has been continuously detected for a certain period of time.

Next, the processing device 1 calculates the reduced area G(t) of the stuck slag candidate that has decreased from the difference F(t-Tt2) a certain period ago to the difference F(t) at time t (S4-7). . Specifically, the reduced area G(t) is determined using the following equation.

G(t)=F(t-Th2) and ~F(t) and(S(t-Th2) and S(t))

Here, and is a logical product, and ~ is an operator that inverts a black and white image. Th2 is a set value.

If the area of the difference F(t) is reduced in a short time due to the effect of slug motion detection, there is a possibility that the slug motion is overdetected. If the motion of the slug is over-detected, it is preferable not to reflect the motion detection results, so in this process, we detect the area reduction of the difference F(t) and decide whether or not to reflect the motion detection results. decide.

Next, the processing device 1 determines whether the ratio of the area of the reduced region G(t) to the area of the slag outlet is smaller than the threshold Th3 (S4-8). If the calculated ratio is smaller than the threshold Th3 (S4-8: YES), the processing device 1 outputs the difference F(t) as the fixed slag area Y(t) (S4-9a). On the other hand, if the calculated ratio is equal to or higher than the threshold Th3 (S4-8: NO), the processing device 1 outputs the common area B(t), which does not take into account the motion detection results, as the fixed slug area Y(t). (S4-9b).

With the above, the fixed slag area Y(t) acquisition process S4 is completed.

Returning to the explanation of FIG. 4. When the fixed slag area acquisition processing S4 is completed, the processing device 1 calculates the aperture ratio of the slag outlet based on the fixed slag area Y(t) (S5: processing as the aperture ratio calculation unit 17). Specifically, the aperture ratio of the slag outlet is determined by the following equation as the area of the fixed slag region Y(t) relative to the pre-registered area of the slag outlet shape when fully open.

A=(HS)/H

Here, A is the opening ratio of the slag outlet, H is the area of the slag outlet shape when fully opened, and S is the area of the fixed slag region.

Through the above procedure, the slag outlet opening ratio is calculated. If the aperture ratio A is smaller than the threshold T, the processing device 1 issues an alarm. Based on this alarm, the operator or controller performs slag removal. This makes it possible to automate the decision to remove slag, which was previously made by humans.

According to the embodiment described above, since fixed slag and molten slag are discriminated by detecting the movement of the molten slag, it is possible to improve the discrimination accuracy. Furthermore, by expanding the movement detection area within the fixed slag candidate area, it is possible to improve the accuracy of detecting the movement of the molten slag.

Although the embodiments of the present invention have been described above, the present invention is not limited to the embodiments described above, and it goes without saying that various modifications can be made by those skilled in the art.

1 processing device, 11 image acquisition section, 12 slag candidate region extraction section, 13 positional deviation correction section, 14 motion detection section, 15 motion detection region correction section, 16 fixed slag region identification section, 17 aperture ratio calculation section, 2 gasification Melting furnace, 21 Melting furnace, 23 Inlet passage, 25 Slag outlet, 27 Discharge passage, 29 Monitoring hole, 3 Camera, 100 Melting furnace monitoring device, 200 Gasification melting furnace

Claims (9)

A camera that continuously photographs the slag outlet of the melting furnace to generate multiple time-series images;
a slug candidate area extraction unit that extracts slug candidate areas in the plurality of images by determination based on a brightness threshold;
a motion detection unit that detects an area in which movement is detected in the plurality of images as a motion detection area;
a stuck slug area identifying unit that identifies an area of the slug candidate area excluding the motion detection area as a stuck slag area;
A melting furnace monitoring system equipped with: further comprising an aperture ratio calculation unit that calculates an aperture ratio of the slag outlet based on the fixed slag area;
The melting furnace monitoring system according to claim 1. The slug candidate region extracting unit extracts, as the slag candidate region, a region commonly extracted in the plurality of images by determination based on a brightness threshold.
The melting furnace monitoring system according to claim 1 or 2. further comprising a positional deviation correction unit that corrects a positional deviation of each of the plurality of images with respect to the immediately preceding image;
The motion detection unit detects the motion detection area in the plurality of images whose positional deviation has been corrected.
A melting furnace monitoring system according to any one of claims 1 to 3. further comprising a motion detection area correction unit that expands the motion detection area within the slug candidate area;
A melting furnace monitoring system according to any one of claims 1 to 4. The motion detection unit detects a region in which motion is detected in any of the plurality of images as the motion detection region.
A melting furnace monitoring system according to any one of claims 1 to 5. The motion detection unit detects the motion detection area by optical flow.
The melting furnace monitoring system according to claim 1. A camera continuously photographs the slag outlet of the melting furnace to generate multiple time-series images.
Extracting slug candidate areas in the plurality of images by determination based on a brightness threshold;
detecting an area in which movement is detected in the plurality of images as a motion detection area;
identifying a region of the slug candidate region excluding the motion detection region as a fixed slag region;
Melting furnace monitoring method. an image acquisition unit that acquires a plurality of time-series images generated by a camera that continuously photographs the slag outlet of the melting furnace;
a slug candidate area extraction unit that extracts slug candidate areas in the plurality of images by determination based on a brightness threshold;
a motion detection unit that detects an area in which movement is detected in the plurality of images as a motion detection area;
a stuck slug area identifying unit that identifies an area of the slug candidate area excluding the motion detection area as a stuck slag area;
A program that allows a computer to function as a computer.

Images